8. Multi-AS Backbones

The same procedures described in Section 10 of [BGP/MPLS-VPN] can be used (and have the same scalability properties) to address the situation where two sites of an IPv6 VPN are connected to different Autonomous Systems. However, some additional points should be noted

when applying these procedures for IPv6 VPNs; these are further described in the remainder of this section.

Approach (a): VRF-to-VRF connections at the AS (Autonomous System) border routers.

This approach is the equivalent for IPv6 VPNs to procedure (a) in Section 10 of [BGP/MPLS-VPN]. In the case of IPv6 VPNs, IPv6 needs to be activated on the inter-ASBR VRF-to-VRF (sub)interfaces. In this approach, the ASBRs exchange IPv6 routes (as opposed to VPN-IPv6 routes) and may peer over IPv6 or over IPv4. The exchange of IPv6 routes MUST be carried out as per [BGP-IPv6]. This method does not use inter-AS LSPs.

Finally, note that with this procedure, since every AS independently implements the intra-AS procedures for IPv6 VPNs described in this document, the participating ASes may all internally use IPv4 tunneling, or IPv6 tunneling; or alternatively, some participating ASes may internally use IPv4 tunneling while others use IPv6 tunneling.

Approach (b): EBGP redistribution of labeled VPN-IPv6 routes from AS to neighboring AS.

This approach is the equivalent for IPv6 VPNs to procedure (b) in Section 10 of [BGP/MPLS-VPN]. With this approach, the ASBRs use EBGP to redistribute labeled VPN-IPv4 routes to ASBRs in other ASes.

In this approach, IPv6 may or may not be activated on the inter-ASBR links since the ASBRs exchanging VPN-IPv6 routes may peer over IPv4 or IPv6 (in which case, IPv6 obviously needs to be activated on the inter-ASBR link). The exchange of labeled VPN-IPv6 routes MUST be carried out as per [BGP-IPv6] and [MPLS-BGP]. When the VPN-IPv6 traffic is to be transported using IPv6 tunneling, the BGP Next Hop Field SHALL contain an IPv6 address. When the VPN-IPv6 traffic is to be transported using IPv4 tunneling, the BGP Next Hop Field SHALL contain an IPv4 address encoded as an IPv4-mapped IPv6 address.

This approach requires that there be inter-AS LSPs. As such, the corresponding (security) considerations described for procedure (b) in Section 10 of [BGP/MPLS-VPN] apply equally to this approach for IPv6.

Finally, note that with this procedure, as with procedure (a), since every AS independently implements the intra-AS procedures for IPv6 VPNs described in this document, the participating ASes may all internally use IPv4 tunneling or IPv6 tunneling; alternatively, some participating ASes may internally use IPv4 tunneling while others use IPv6 tunneling.

Approach (c): Multihop EBGP redistribution of labeled VPN-IPv6 routes between source and destination ASes, with EBGP redistribution of labeled IPv4 or IPv6 routes from AS to neighboring AS.

This approach is equivalent for exchange of VPN-IPv6 routes to procedure (c) in Section 10 of [BGP/MPLS-VPN] for exchange of VPN-IPv4 routes.

This approach requires that the participating ASes either all use IPv4 tunneling or all use IPv6 tunneling.

In this approach, VPN-IPv6 routes are neither maintained nor distributed by the ASBR routers. The ASBR routers need not be dual stack. An ASBR needs to maintain labeled IPv4 (or IPv6) routes to the PE routers within its AS. It uses EBGP to distribute these routes to other ASes. ASBRs in any transit ASes will also have to use EBGP to pass along the labeled IPv4 (or IPv6) routes. This results in the creation of an IPv4 (or IPv6) label switch path from ingress PE router to egress PE router. Now, PE routers in different ASes can establish multi-hop EBGP connections to each other over IPv4 or IPv6 and can exchange labeled VPN-IPv6 routes over those EBGP connections. Note that the BGP Next Hop field of these distributed VPN-IPv6 routes will contain an IPv6 address when IPv6 tunneling is used or an IPv4-mapped IPv6 address when IPv4 tunneling is used.

The considerations described for procedure (c) in Section 10 of [BGP/MPLS-VPN] with respect to possible use of route-reflectors, with respect to possible use of a third label, and with respect to LSPs spanning multiple ASes apply equally to this IPv6 VPN approach.